Current-Users archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: panic! panic! panic! sigh.. panic!

btw, just to follow up on this one:

On Sun, Sep 21, 2008 at 02:25:04AM +0200, Quentin Garnier wrote:
> On Sat, Sep 20, 2008 at 07:58:02PM -0400, wrote:
> > (...)
> > -makeoptions    CPUFLAGS="-march=i486 -mtune=pentiumpro" # Assume at least 
> > 486, ...
> > +makeoptions    CPUFLAGS="-march=pentium4 -msse2 -mfpmath=sse" 
> You really don't want the kernel to use any FP instruction.  You
> shouldn't have the sse flags on.

First of all, that might have been it -- no hangup/panic for now after a night
of building and shoveling stuff around. I'll try with ahci enabled back soon to
see whether it really were my makeoptions which broke the system or the ahci.

about the makeoptions:

I would expect the kernel to not *do* any FP stuff. I thought SSE(2) would
not hurt as I thought it also enables the use of SIMD Integer operations
along with the MMX registers but I'm no expert. So from that baseline of
thought you'd welcome sse in the kernel.

Furthermore, with devel/cpuflags, which also sets CPUFLAGS, I kind of
made the link to the makeoptions in the kernel, assuming that whetever
cpuflags(1) is outputting I can use also in the kernel for makeoptions CPUFLAGS.

Maybe the fact that this is not true should be documented somewhere.



Home | Main Index | Thread Index | Old Index